The use of tumor immunotherapy for solid tumors has expanded dramatically with the development of immune checkpoint inhibitors, especially antibodies to PD-1/PD-L1 that are approved across many cancers. Yet, even in melanoma, where single-agent PD1 inhibition has the greatest response rates (40-45%), the majority of patients will not respond and ultimately succumb to disease. Identification of mechanisms of intrinsic and acquired resistance to anti-PD1 therapy remains the greatest unmet need in the field of tumor immunology and perhaps medical oncology. The elucidation of how tumors resist the immune system in the setting of PD-1 inhibition, and the development of strategies to target these mechanisms of resistance stands to change the way cancer patients are treated. We have recently discovered that ~30% of melanoma patients harbor tumorspecific mutations in the beta-2-microglobulin (B2M) gene, which is essential for presenting antigens to T cells for direct tumor recognition and killing. In two independent cohorts of 105 and 38 melanoma patients treated with anti-CTLA4 and anti-PD1, respectively, we found deletions of B2M to be 3-fold enriched in nonresponders (30%) vs. responders (10%) and significantly associated with lower overall survival -- with homozygote loss of B2M found only in non-responders. These results imply that B2M loss is a fundamental mechanism of intrinsic and acquired resistance to checkpoint inhibitors. A surprising result was that even in some of the tumors with single B2M deletions there was little or no expression of the B2M protein in tumor cells by IHC. In addition, we hypothesized and found that NK cells play a role in controlling growth of B2M-deficient tumors in a mouse model of melanoma that is deficient in B2M. In this proposal, we will analyze the evolution of B2M mutations in melanoma and NSCLC patients receiving anti-PD1 therapy and thus create a method for real time tracking of resistance to checkpoint blockade therapies. Furthermore, we will test the role of epigenetic silencing of B2M to explain the loss of its expression in tumors missing one copy of the gene, and will attempt to restore B2M expression by modulating epigenetic regulators. Since a central goal is to overcome resistance, we will activate natural killer (NK) cells in an attempt to kill cells lacking B2M in mouse models of melanoma. In all the studies, we will use a combination of agents that include ones from CTEP. By monitoring, explaining and hopefully overcoming resistance to checkpoint therapy by B2M loss, our research will help provide a path to address this fundamental form of resistance to checkpoint blockade